Towards the automation of a defect detection protocol for functional size measurements

Abstract

There is a need to develop formal protocols to verify the accuracy of software functional size measurements and to calibrate the measurement processes. These protocols offer a mechanism for accomplishing such verification regardless of who performed the measurements (an expert or an automated tool). The detailed analysis of functional size measurement procedures results is necessary to understand the nature of the defects. The automation of these protocols decreases the effort and time required to verify the measurements accuracy and supports the calibration of procedures and tools for functional size measurement. This paper presents an empirical study for evaluating the results of a prototype tool that automates an accuracy verification protocol for IFPUG FPA measurements. Our tool uses a graph-based model to search for the causes of measurement mismatches detected in the verification process. The current prototype has reached an accuracy of 93.92%, a precision of 98.69% and a recall of 94.94%.